Fast Inverse Planning of Beam Directions and Weights for Small Animal Radiotherapy

Nathan B. Cho, John Wong, Peter Kazanzides

Research output: Contribution to journalArticlepeer-review


Current preclinical irradiation systems, such as the small animal radiation research platform, provide many capabilities of clinical systems, including on-board cone beam computed tomography (CBCT) imaging, which motivates the development of treatment planning systems. The time for preclinical treatment planning is often limited, however, due to the large number of subjects and the need for anesthesia for imaging and radiation delivery. This paper presents a 3-D inverse planning solution that optimizes beam directions and weights within a timeframe suitable for small animal radiation research. The system begins with a large number of beams and takes advantage of a graphics processing unit-accelerated implementation of the superposition-convolution method to quickly compute the dose for each beam. Optimization is performed by a linear programming solver, with a minimum bound on dose to the target and a maximum bound on dose to organs at risk (OAR). It is extended to dose shells by using hollow cylinders instead of solid beams. The method is demonstrated on mouse CBCT images with arbitrarily defined targets and OAR. Inverse planning is performed in about 5 min, which enables a preclinical workflow that includes CBCT acquisition, treatment planning, and radiation delivery in a single session.

Original languageEnglish (US)
Pages (from-to)215-222
Number of pages8
JournalIEEE Transactions on Radiation and Plasma Medical Sciences
Issue number3
StatePublished - May 2018


  • Inverse treatment planning
  • preclinical radiation research
  • radiotherapy
  • small animal

ASJC Scopus subject areas

  • Atomic and Molecular Physics, and Optics
  • Radiology Nuclear Medicine and imaging
  • Instrumentation


Dive into the research topics of 'Fast Inverse Planning of Beam Directions and Weights for Small Animal Radiotherapy'. Together they form a unique fingerprint.

Cite this